WO2006124414A2 - Queue aware flow control - Google Patents

Queue aware flow control Download PDF

Info

Publication number
WO2006124414A2
WO2006124414A2 PCT/US2006/017923 US2006017923W WO2006124414A2 WO 2006124414 A2 WO2006124414 A2 WO 2006124414A2 US 2006017923 W US2006017923 W US 2006017923W WO 2006124414 A2 WO2006124414 A2 WO 2006124414A2
Authority
WO
WIPO (PCT)
Prior art keywords
pause
flow
transmitter
receiver
aware
Prior art date
Application number
PCT/US2006/017923
Other languages
French (fr)
Other versions
WO2006124414A3 (en
Inventor
Ankur Singla
Harshad Nakil
Original Assignee
Cisco Technology, Inc.
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Cisco Technology, Inc. filed Critical Cisco Technology, Inc.
Priority to EP06752446A priority Critical patent/EP1882336B1/en
Priority to CN200680010068.1A priority patent/CN101536413B/en
Publication of WO2006124414A2 publication Critical patent/WO2006124414A2/en
Publication of WO2006124414A3 publication Critical patent/WO2006124414A3/en

Links

Classifications

    • HELECTRICITY
    • H04ELECTRIC COMMUNICATION TECHNIQUE
    • H04LTRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L47/00Traffic control in data switching networks
    • H04L47/10Flow control; Congestion control
    • HELECTRICITY
    • H04ELECTRIC COMMUNICATION TECHNIQUE
    • H04LTRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L47/00Traffic control in data switching networks
    • H04L47/50Queue scheduling
    • HELECTRICITY
    • H04ELECTRIC COMMUNICATION TECHNIQUE
    • H04LTRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L47/00Traffic control in data switching networks
    • H04L47/50Queue scheduling
    • H04L47/62Queue scheduling characterised by scheduling criteria
    • H04L47/6205Arrangements for avoiding head of line blocking
    • HELECTRICITY
    • H04ELECTRIC COMMUNICATION TECHNIQUE
    • H04LTRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L49/00Packet switching elements
    • H04L49/90Buffering arrangements
    • H04L49/9063Intermediate storage in different physical parts of a node or terminal
    • H04L49/9078Intermediate storage in different physical parts of a node or terminal using an external memory or storage device

Definitions

  • a router or switch directs packets from an input to an output.
  • the classic network device architecture includes multiple line cards having onboard queue managers, that connect to a central switching mechanism.
  • Fig. 1 is a high level depiction of a network device 10 depicting line cards 12 connected to a crossbar switch 14 by serial links 16.
  • the crossbar includes crossbar switches in the form of ASICs coupled to a backplane formed of multiple serial links.
  • the line cards act as the inputs and outputs of the router and are coupled to a backplane switching fabric which performs the task of coupling an input line card to an output line card for data transfer. Traffic from the input line card is passed through a switch the backplane switching fabric and passed to the proper output line card.
  • High speed network device applications utilize a serial backplane where data is transferred on high-speed serial paths formed in the backplane.
  • the queue managers must implement flow control to prevent data from being lost when buffers on a receiver overflow. Backpressure is used to control flow and the receiver has the ability to prevent transmission of data until its buffers have capacity to receive more data.
  • head of line (HOL) blocking may occur when a packet at the head of the transmit queue can't be transmitted because of backpressure. This blocks all packets in the transmit queue from being transmitted even if their intended receiver is available.
  • VOQs virtual output queues
  • Flow Control is a process that is found in most Ethernet networks. It is needed to ensure that devices do not overload other devices on the network. Special frames called 'PAUSE frames' are used to signal traffic flow requests and status between nodes. PAUSE frames permit one end station to temporarily stop all traffic from the other end station (except MAC Control frames).
  • Station A transmits frames at a rate that causes Station B to enter into a state of congestion (i.e. no buffer space remaining to receive additional frames).
  • Station B may transmit a PAUSE frame to Station A requesting that Station A stop transmitting frames for a specified period of time.
  • Station A Upon receiving the PAUSE frame, Station A will suspend further frame transmission until the specified time period has elapsed. This will allow Station B time to recover from the congestion state. At the end of the specified time period, Station A will resume normal transmission of frames.
  • PAUSE frame protocol is bi-directional. Station A may send frames to pause Station B, and Station B may send frames to pause Station A.
  • a PAUSE frame is the one type of frame that a station is allowed to send even if it is currently in the paused state.
  • Support for PAUSE frames is optional among devices that implement the full- duplex protocol (the use of PAUSE frames is not supported in a half-duplex environment). It is valid for a device to support only half of the protocol; i.e. it may transmit PAUSE frames without having the capability to decode them on the receive side, and vice-versa.
  • Devices use the Auto-Negotiation protocol to discover the PAUSE frame capabilities of the device at the other end of the link. This permits interoperability between devices that do or do not support one or both halves of the protocol.
  • FIG. 1 is a block diagram of an example the architecture of a network device
  • FIG. 2 is a block diagram of a conventional MAC
  • FIG. 3 is block diagram of an embodiment of the invention.
  • FIGs. 4 and 5 are flow charts of actions performed by embodiments of the invention.
  • Ethernet Protocol MultiGigabit SERDES backplane for Access Routing platforms that utilize Ethernet Protocol.
  • the use of Ethernet Protocol will help alleviate the congestion of the currently utilized PCI- based backplane and provide a low cost and low power point-to-point switched interconnect.
  • the current Ethernet Flow Control protocol does not provide the required support for backplane applications.
  • backplanes carry traffic in many virtual streams bound for different destinations.
  • Datagram traffic (latency insensitive) and Control traffic (latency sensitive) are examples of two virtual streams in a system.
  • the incoming virtual stream at that input ⁇ ort(s) of the switching node may enter a blocked state as it fills up.
  • MAC Ethernet Media Access Controller
  • Fig. 2 depicts the existing MAC system utilized in Ethernet systems in compliance with IEEE 802.3.
  • the system includes Tx and Rx FIFOs, Pause Frame generation logic (PFGL), an 802.3 processing element, pause frame handling logic (PFFIL) and a Physical Layer (PHY) device coupled to the media.
  • PFGL Pause Frame generation logic
  • PFFIL pause frame handling logic
  • PHY Physical Layer
  • the system logic e.g., a DMA or switch scheduler
  • the 802.3 processing element receives frames into the Rx FIFO.
  • the pause frame generation logic When the fill level of the Rx FIFO is above a programmable threshold the pause frame generation logic generates a pause frame with a programmable pause value and passes the pause frame to the 802.3 processing element for transmission to the PHY. When the fill level of the Rx FIFO is below another programmable level the pause frame generation logic generates a pause frame with a zero pause value to signal a peer device to resume transmission.
  • the Rx pause frame logic informs the 802.3 processing element to stop transmission for the pause time after completing transmission of the current frame.
  • the Rx pause frame logic informs the 802.3 processing element to resume transfers when a pause frame with a zero pause is received.
  • the standard 802.3x pause frame format included a Length/Type field of 16 bits that holds a value 88-08 (hex) to indicate a MAC control frame, a MAC Control Opcode of 16 bits that holds a value 00-01 to indicate that the frame is being used as a Pause Frame, and a Pausejime parameter field of 16 bits. Additionally, the Pause Frame includes a 42-byte reserved field (transmitted as all zeros) that is required to pad the length of the Pause Frame to the minimum Ethernet frame size.
  • Fig 3 is a block diagram of an embodiment of the invention.
  • System logic is coupled to sets of multiple Tx and Rx FIFOs.
  • a scheduler has inputs coupled to Tx FIFOs and an output coupled to the 802.3 processing element.
  • Pause Frame generation logic has control inputs coupled to the Rx FIFOs and a pause frame output coupled to the 802.3 processing element and
  • Pause Frame handling logic has a Pause Frame input coupled to the 802.3 processing element and a control output coupled to the scheduler.
  • the 802.3 processing element has Tx and Rx ports coupled to the PHY.
  • the reserved field of the standard 802.3x Pause Frame is utilized to identify a particular virtual stream to be paused.
  • This information can be encoded by assigning a bit position in the reserved field to each virtual stream, e.g., a particular Rx FIFO in a receiver and Tx FIFO in a transmitter that are exchanging information, and the setting of a bit in one of the bit positions identifies the particular Tx FIFO in the set of Tx FIFOs that is to be paused.
  • a Flow- A ware Pause Frame is generated encoding the programmable pause time and having the bit position identifying Rx FIFO 1 in the reserved field of the Pause Frame set by the Pause Frame generation logic.
  • the FAPF is then passed to the 802.3 processing element for transmission.
  • a FAPF when received by the 802.3 processing element, it is passed to the PFHL which reads the pause time and virtual stream identification information from the Pause Frame.
  • the pause time and virtual stream identification information are sent from the PFHL to the Scheduler.
  • the Scheduler stops scheduling the identified Tx FIFO for transmission for the amount of time indicated by the pause time.
  • an Ethernet device determines that the link partner does not support this modification, it will default to standard Ethernet flow control mechanism. This allows for support for off-the-shelf Ethernet devices.
  • TX Standard MAC - RX Standard MAC -> HOL blocking can occur. No traffic class differentiation.
  • TX Standard MAC - RX enhanced MAC -> HOL blocking can occur. No traffic class differentiation.
  • TX enhanced MAC - RX standard MAC -> HOL blocking can occur. No traffic class differentiation.
  • SERDES, SGMII and/or TBI reserved K-codes in the 8b/10b encoded PMD/PCS sub-layer are used to interrupt the packet being transmitted and insert the modified standard PAUSE frame or a specially defined PAUSE frame.
  • a description of 8b/10b encoding in K codes is given in an article by Z.X. Widmer and P. A. Franaszek entitled A DC-BALANCED, PARTITIONED-BLOCK, 8B/10B TRANSMISSION CODE, IBM Journal of Research and Development, Vol. 27, Number 5, September 1983.
  • the flow control message can be inserted by the receiver and received by the transmitter as soon as congestion is detected at the receiver, that is there is no need to wait for frame boundary, the buffering requirements per virtual stream can be reduced thereby reducing the turnaround time from the transmission of the pause frame to actual pause of the transmission from the virtual stream.
  • the Pause Frame Generation Logic triggers the PHY to interrupt the packet being transmitted and to transmit a K-code that indicates to the transmitter that the current frame is being aborted and that the next packets will comprise a PAUSE frame. Accordingly, the delay between the time the RxFIFO overflows and the time that the transmitter pauses the transmission of packets on the virtual stream is reduced.

Abstract

A network flow control system utilizes flow-aware pause frames that identify a specific virtual stream to pause. Special codes may be utilized to interrupt a frame being transmitted to insert a pause frame without waiting for frame boundaries.

Description

PATENT APPLICATION
QUEUE AWARE FLOW CONTROL
BACKGROUND OF THE INVENTION
[01] A router or switch (network device) directs packets from an input to an output. The classic network device architecture includes multiple line cards having onboard queue managers, that connect to a central switching mechanism. Fig. 1 is a high level depiction of a network device 10 depicting line cards 12 connected to a crossbar switch 14 by serial links 16. The crossbar includes crossbar switches in the form of ASICs coupled to a backplane formed of multiple serial links. The line cards act as the inputs and outputs of the router and are coupled to a backplane switching fabric which performs the task of coupling an input line card to an output line card for data transfer. Traffic from the input line card is passed through a switch the backplane switching fabric and passed to the proper output line card. High speed network device applications utilize a serial backplane where data is transferred on high-speed serial paths formed in the backplane.
[02] The queue managers must implement flow control to prevent data from being lost when buffers on a receiver overflow. Backpressure is used to control flow and the receiver has the ability to prevent transmission of data until its buffers have capacity to receive more data.
[03] At the transmitter, head of line (HOL) blocking may occur when a packet at the head of the transmit queue can't be transmitted because of backpressure. This blocks all packets in the transmit queue from being transmitted even if their intended receiver is available.
[04] The problem of HOL blocking is solved by forming virtual output queues (VOQs) on the line cards where a VOQ is established for each potential receiver. Thus, if the head packet on a VOQ is blocked the line card can switch to another VOQ having a head packet that is not blocked.
[OS] The backplane interconnect architecture is a key chokepoint in today's communication infrastructure designs. A great deal of attention has been paid to moving data between chips on a board and there has been growing acceptance of Gigabit and Multi- Gigabit Ethernet as a point to point switched interconnect technology.
[06] Flow Control is a process that is found in most Ethernet networks. It is needed to ensure that devices do not overload other devices on the network. Special frames called 'PAUSE frames' are used to signal traffic flow requests and status between nodes. PAUSE frames permit one end station to temporarily stop all traffic from the other end station (except MAC Control frames).
[07] For example, assume a full-duplex link that connects two devices called "Station A" and "Station B". Suppose Station A transmits frames at a rate that causes Station B to enter into a state of congestion (i.e. no buffer space remaining to receive additional frames). Station B may transmit a PAUSE frame to Station A requesting that Station A stop transmitting frames for a specified period of time. Upon receiving the PAUSE frame, Station A will suspend further frame transmission until the specified time period has elapsed. This will allow Station B time to recover from the congestion state. At the end of the specified time period, Station A will resume normal transmission of frames.
[08] Note that the PAUSE frame protocol is bi-directional. Station A may send frames to pause Station B, and Station B may send frames to pause Station A. A PAUSE frame is the one type of frame that a station is allowed to send even if it is currently in the paused state. Support for PAUSE frames is optional among devices that implement the full- duplex protocol (the use of PAUSE frames is not supported in a half-duplex environment). It is valid for a device to support only half of the protocol; i.e. it may transmit PAUSE frames without having the capability to decode them on the receive side, and vice-versa. Devices use the Auto-Negotiation protocol to discover the PAUSE frame capabilities of the device at the other end of the link. This permits interoperability between devices that do or do not support one or both halves of the protocol.
[09] However, the current flow control using PAUSE frame does not fit very well in the backplane architecture as it does not distinguish internal queues of the system and prevents traffic to flow between unclogged queues when one of the queues become full.
[10] The problem with PAUSE frames is that the transmitting device stops all further transmissions for duration of the PAUSE event in units of 512-bit times. This works very well for devices connected over the Ethernet network in a LAN/WAN environment. However, this can become a Quality of Service (QoS) bottleneck in a backplane interconnect environment.
[11] The challenges in the field of backplane interconnect design continue to increase with demands for more and better techniques having greater flexibility and .adaptability. Therefore, a need has arisen for a new system and method for providing backpressure control in Gigabit Ethernet systems allowing for QoS and Class of Service (CoS) requirements to be met. BRIEF DESCRIPTION OF THE DRAWINGS
[12] Fig. 1 is a block diagram of an example the architecture of a network device;
[13] Fig. 2 is a block diagram of a conventional MAC;
[14] Fig. 3 is block diagram of an embodiment of the invention; and
[15] Figs. 4 and 5 are flow charts of actions performed by embodiments of the invention.
DETAILED DESCRIPTION OF THE INVENTION [16] Reference will now be made in detail to various embodiments of the invention. Examples of these embodiments are illustrated in the accompanying drawings. While the invention will be described in conjunction with these embodiments, it will be understood that it is not intended to limit the invention to any embodiment. On the contrary, it is intended to cover alternatives, modifications, and equivalents as may be included within the spirit and scope of the invention as defined by the appended claims. In the following description, numerous specific details are set forth in order to provide a thorough understanding of the various embodiments. However, the present invention maybe practiced without some or all of these specific details. La other instances, well known process operations have not been described in detail in order not to unnecessarily obscure the present invention.
[17] A first embodiment will be described in connection with a
MultiGigabit SERDES backplane for Access Routing platforms that utilize Ethernet Protocol. The use of Ethernet Protocol will help alleviate the congestion of the currently utilized PCI- based backplane and provide a low cost and low power point-to-point switched interconnect. However, the current Ethernet Flow Control protocol does not provide the required support for backplane applications.
[18] Typically, backplanes carry traffic in many virtual streams bound for different destinations. Datagram traffic (latency insensitive) and Control traffic (latency sensitive) are examples of two virtual streams in a system. When one of the outgoing line- cards has one of its virtual streams full, the incoming virtual stream at that input ρort(s) of the switching node may enter a blocked state as it fills up. If a current Ethernet Media Access Controller (MAC) is used, it can cause all traffic from many line cards to stop flowing to the switching node as one of the virtual streams is blocked. This is not desirable in the backplane interconnect. [19] Even during heavy loading of the switched backplane, it should be possible to transport latency sensitive traffic and/or control traffic without getting head of line blocked because the traffic of one of the virtual streams is blocked. The avoidance of HOL blocking requires virtual stream aware flow control support in the Ethernet protocol.
[20] Ba the following, two techniques for providing virtual stream aware flow control in the Ethernet protocol are described. Both of these techniques are novel in a way that they not only support standard Ethernet MACs but also allow for a mixed fabric of modified and standard Ethernet devices. This is especially important in access routing market where time to market dictates development of modules/line cards using off-the-shelf components,
[21] Fig. 2 depicts the existing MAC system utilized in Ethernet systems in compliance with IEEE 802.3. The system includes Tx and Rx FIFOs, Pause Frame generation logic (PFGL), an 802.3 processing element, pause frame handling logic (PFFIL) and a Physical Layer (PHY) device coupled to the media.
[22] When transmitting, the system logic, e.g., a DMA or switch scheduler, puts frames for transmission in the Tx FIFO and 802.3 processing element transmits frames from the Tx FIFO to the PHY. When receiving, the 802.3 processing element receives frames into the Rx FIFO.
[23] When the fill level of the Rx FIFO is above a programmable threshold the pause frame generation logic generates a pause frame with a programmable pause value and passes the pause frame to the 802.3 processing element for transmission to the PHY. When the fill level of the Rx FIFO is below another programmable level the pause frame generation logic generates a pause frame with a zero pause value to signal a peer device to resume transmission.
[24] When a pause frame with a non-zero pause is received, the Rx pause frame logic informs the 802.3 processing element to stop transmission for the pause time after completing transmission of the current frame. The Rx pause frame logic informs the 802.3 processing element to resume transfers when a pause frame with a zero pause is received.
[25] The standard 802.3x pause frame format included a Length/Type field of 16 bits that holds a value 88-08 (hex) to indicate a MAC control frame, a MAC Control Opcode of 16 bits that holds a value 00-01 to indicate that the frame is being used as a Pause Frame, and a Pausejime parameter field of 16 bits. Additionally, the Pause Frame includes a 42-byte reserved field (transmitted as all zeros) that is required to pad the length of the Pause Frame to the minimum Ethernet frame size.
[26] Fig 3 is a block diagram of an embodiment of the invention. System logic is coupled to sets of multiple Tx and Rx FIFOs. A scheduler has inputs coupled to Tx FIFOs and an output coupled to the 802.3 processing element. Pause Frame generation logic has control inputs coupled to the Rx FIFOs and a pause frame output coupled to the 802.3 processing element and Pause Frame handling logic has a Pause Frame input coupled to the 802.3 processing element and a control output coupled to the scheduler. The 802.3 processing element has Tx and Rx ports coupled to the PHY.
[27] In this embodiment, the reserved field of the standard 802.3x Pause Frame is utilized to identify a particular virtual stream to be paused. Two bytes are used to define the virtual stream map [16 virtual streams] and two bytes define the spaces left for max sized messages per virtual stream [total of 16x2 = 32 bytes]. Rest of the 8 bytes are reserved for future use.
[28] This information can be encoded by assigning a bit position in the reserved field to each virtual stream, e.g., a particular Rx FIFO in a receiver and Tx FIFO in a transmitter that are exchanging information, and the setting of a bit in one of the bit positions identifies the particular Tx FIFO in the set of Tx FIFOs that is to be paused.
[29] The operation of the embodiment depicted in Fig. 3 will now be described. The operation of the PFGL to generate a flow-aware pause frame (FAPF) will be described with reference to Fig. 4 and the operation of the PFHL to handle received FAPF will be described with reference to Fig. 5.
. [30] In Fig. 4, if the fill level of a particular Rx FIFO, in this example Rx FIFO 1, exceeds the programmable threshold then a Flow- A ware Pause Frame is generated encoding the programmable pause time and having the bit position identifying Rx FIFO 1 in the reserved field of the Pause Frame set by the Pause Frame generation logic. The FAPF is then passed to the 802.3 processing element for transmission.
[31] In Fig. 5, when a FAPF is received by the 802.3 processing element, it is passed to the PFHL which reads the pause time and virtual stream identification information from the Pause Frame. The pause time and virtual stream identification information are sent from the PFHL to the Scheduler. The Scheduler stops scheduling the identified Tx FIFO for transmission for the amount of time indicated by the pause time.
[32] If during negotiation, either Link Level Auto negotiation or software initiated negotiation, an Ethernet device determines that the link partner does not support this modification, it will default to standard Ethernet flow control mechanism. This allows for support for off-the-shelf Ethernet devices.
[33] Overall, 4 working configurations and their performance would be:
1. TX Standard MAC - RX Standard MAC -> HOL blocking can occur. No traffic class differentiation.
2. TX Standard MAC - RX enhanced MAC -> HOL blocking can occur. No traffic class differentiation.
3. TX enhanced MAC - RX standard MAC -> HOL blocking can occur. No traffic class differentiation.
4. TX enhanced MAC - RX enhanced MAC -> Full traffic class differentiation.
[34] However, when all link partners use Enhanced MAC, there are significant benefits to overall architecture as congestion information of one flow is fed back from the congestion point to the sources and helps relieve congestion by selectively backing- off only sources belonging to a particular virtual stream. The Pause Frame Handling Logic at the transmitter decodes the identity of the virtual stream requesting a pause and the scheduler PAUSEes all TxFIFOs belonging to the identified virtual stream.
[35] In another embodiment of the invention utilizes Reserved Code Words to interrupt a frame being transmitted to immediately transmit a Pause Frame. In a standard Ethernet system the transmission of the Pause Frame is delayed until the transmission of the current frame is completed. This could lead to overflow of the Rx buffer that requested the Pause Frame.
[36] In this embodiment, SERDES, SGMII and/or TBI reserved K-codes in the 8b/10b encoded PMD/PCS sub-layer are used to interrupt the packet being transmitted and insert the modified standard PAUSE frame or a specially defined PAUSE frame. A description of 8b/10b encoding in K codes is given in an article by Z.X. Widmer and P. A. Franaszek entitled A DC-BALANCED, PARTITIONED-BLOCK, 8B/10B TRANSMISSION CODE, IBM Journal of Research and Development, Vol. 27, Number 5, September 1983.
[37] Since the flow control message can be inserted by the receiver and received by the transmitter as soon as congestion is detected at the receiver, that is there is no need to wait for frame boundary, the buffering requirements per virtual stream can be reduced thereby reducing the turnaround time from the transmission of the pause frame to actual pause of the transmission from the virtual stream.
[38] Referring back to Fig. 3, when an RxFIFO overflows the Pause Frame Generation Logic triggers the PHY to interrupt the packet being transmitted and to transmit a K-code that indicates to the transmitter that the current frame is being aborted and that the next packets will comprise a PAUSE frame. Accordingly, the delay between the time the RxFIFO overflows and the time that the transmitter pauses the transmission of packets on the virtual stream is reduced.
[39] The invention has now been described with reference to the preferred embodiments. Alternatives and substitutions will now be apparent to persons of skill in the art. For example, as is known in the art, the FIFO overflow condition can be defined in many ways. Further, virtual streams experiencing backpressure can be identified using different combinations of bit positions in the reserved fields of the Pause Frame. Also, the invention is not limited to systems in strict compliance with BEEE 802.3. Accordingly, it is not intended to limit the invention except as provided by the appended claims.

Claims

WHAT IS CLAMED IS:
1. A method for controlling flow between a transmitter/receiver pair, with the receiver having a plurality of Rx FIFOs holding frames received at the receiver and the transmitter having a plurality of TxFIFOs, the method comprising: generating an overflow indication identifying an overflow RxFIFO having a fill level exceeding a threshold; forming a flow-aware pause frame indicating the identity of the overflow RxFIFO and indicating a pause time value; transmitting the flow-aware pause frame to the transmitter; receiving the flow-aware pause frame at the transmitter and decoding the flow-aware pause frame to determine the identity of a TxFIFO that transmits packets to the overflow RxFIFO and the pause time value; and stopping transmission of frames from an identified TxFIFO to the overflow RxFIFO for the pause time value.
2. The method of claim 1 further comprising: utilizing reserved bits in flow-aware pause frame to indicate the identity of the overflow RxFBFO.
3. The method of claim 1 further comprising: auto-negotiating between the transmitter and receiver to determine whether flow-aware pause frames are supported.
4. A system for controlling flow between a transmitter/receiver pair, said system comprising: a plurality of RxFIFOs at the receiver for holding received frames; a pause frame generation logic block that generates a flow-aware pause frame, when an overflow indication is received, identifying an overflow RxFIFO, with the pause frame generation logic encoding the identity of the overflow RxFIFO in the flow-aware pause frame; a first processing unit on the receiver, coupled to pause frame generation logic block, that transmits the flow-aware pause frame; a second processing on the transmitter for receiving the flow-aware pause frame; a plurality TxFIFOs at the transmitter that hold frames to be transmitted; a scheduler, coupled to the first and second TxFIFOs, that schedules a frame to be transmitted from the TxFIFOs; and a pause frame handling logic block, coupled to the second processor and the scheduler, that decodes a received flow-aware pause frame to identify a TxFIFO that transmits frames to the overflow RxPIFO indicated by the flow-aware pause frame and that signals the scheduler to stop transmitting frames from an identified TxFBFO to the overflow RxFIFO.
5. A method for controlling flow between a transmitter/receiver pair, with the receiver having an Rx FIFO holding packets to be transmitted, the method comprising: generating a full indication at the RxFJOFO when the fill level of the RxFlFO exceeds a threshold; forming, in response to the full indication, a pause frame indicating a pause time value; interrupting a frame being transmitted from the receiver and transmitting a special code in response to pause frame formation; receiving the special code at the transmitter and responding to subsequent packets as a pause frame; and stopping transmission of frames from the first TxFIFO for the pause time value.
6. The method of claim 5 further comprising: autonegotiating to determine whether the transmitter and receiver support use of the special code.
7. A system for controlling flow between a transmitter/receiver pair, with the receiver having a plurality of Rx FIFOs holding frames received at the receiver and the transmitter having a plurality of TxFBFOs, the system comprising: means for generating an overflow indication identifying an overflow RxFIFO having a fill level exceeding a threshold; means for forming a flow-aware pause frame indicating the identity of the overflow RxFIFO and indicating a pause time value; means for transmitting the flow-aware pause frame to the transmitter; means for receiving the flow-aware pause frame at the transmitter and decoding the flow-aware pause frame to determine the identity of a TxFIFO that transmits packets to the overflow RxFIFO and the pause time value; and means for stopping transmission of frames from an identified TxFIFO to the overflow RxFIFO for the pause time value.
8. The system of claim 7 further comprising: means for utilizing reserved bits in flow-aware pause frame to indicate the identity of the overflow RxFIFO.
9. The method of claim 7 further comprising: auto-negotiating between the transmitter and receiver to determine whether flow-aware pause frames are supported.
10. A system for controlling flow between a transmitter/receiver pair, with the receiver having an Rx FIFO holding packets to be transmitted, the sysstem comprising: means for generating a full indication at the RxFIFO when the fill level of the RxFIFO exceeds a threshold; means for forming, in response to the full indication, a pause frame indicating a pause time value; means for interrupting a frame being transmitted from the receiver and transmitting a special code in response to pause frame formation; means for receiving the special code at the transmitter and responding to subsequent packets as a pause frame; means for stopping transmission of frames from the first TxFIFO for the pause time value.
11. The system of claim 10 further comprising: means for autonegotiating to determine whether the transmitter and receiver support use of the special code.
12. A system for controlling flow between a transmitter/receiver pair, with the receiver having a plurality of Rx FIFOs holding frames received at the receiver and the transmitter having a plurality of TxFIFOs, the receiver comprising: means for generating an overflow indication identifying an overflow RxFIFO having a fill level exceeding a threshold; means for forming a flow-aware pause frame indicating the identity of the overflow RxFIFO and indicating a pause time value; means for transmitting the flow-aware pause frame to the transmitter.
13. A system for controlling flow between a transmitter/receiver pair, with the receiver having a plurality of Rx FIFOs holding frames received at the receiver and the transmitter having a plurality of TxFIFOs, the receiver comprising: means for receiving a flow-aware pause frame at the transmitter and decoding the flow-aware pause frame to determine the identity of a TxFIFO that transmits packets to an overflow RxFIFO identified by the flow-aware pause frame and the pause time value; and means for stopping transmission of frames from an identified TxFEFO to the overflow RxFIFO for the pause time value.
PCT/US2006/017923 2005-05-16 2006-05-10 Queue aware flow control WO2006124414A2 (en)

Priority Applications (2)

Application Number Priority Date Filing Date Title
EP06752446A EP1882336B1 (en) 2005-05-16 2006-05-10 Queue aware flow control
CN200680010068.1A CN101536413B (en) 2005-05-16 2006-05-10 Queue aware flow control

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
US11/130,700 2005-05-16
US11/130,700 US7839779B2 (en) 2005-05-16 2005-05-16 Queue aware flow control

Publications (2)

Publication Number Publication Date
WO2006124414A2 true WO2006124414A2 (en) 2006-11-23
WO2006124414A3 WO2006124414A3 (en) 2007-11-22

Family

ID=37431830

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/US2006/017923 WO2006124414A2 (en) 2005-05-16 2006-05-10 Queue aware flow control

Country Status (4)

Country Link
US (1) US7839779B2 (en)
EP (1) EP1882336B1 (en)
CN (1) CN101536413B (en)
WO (1) WO2006124414A2 (en)

Cited By (11)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
EP2048831A1 (en) * 2007-10-12 2009-04-15 Nokia Siemens Networks S.p.A. Method and systems for QoS-aware flow control in communication networks, corresponding network and computer program product
EP2256640A1 (en) * 2009-05-29 2010-12-01 Oracle America, Inc. Managing traffic on virtualized lanes between a network switch and a virtual machine
EP2302847A1 (en) 2009-09-25 2011-03-30 Alcatel Lucent Control of traffic flows with different CoS in a communication network
US8767752B1 (en) 2010-05-03 2014-07-01 Pluribus Networks, Inc. Method and system for resource coherency and analysis in a network
US8891543B1 (en) 2011-05-23 2014-11-18 Pluribus Networks Inc. Method and system for processing packets in a network device
US9154445B2 (en) 2010-05-03 2015-10-06 Pluribus Networks Inc. Servers, switches, and systems with virtual interface to external network connecting hardware and integrated networking driver
US9160668B2 (en) 2010-05-03 2015-10-13 Pluribus Networks Inc. Servers, switches, and systems with switching module implementing a distributed network operating system
US9300576B2 (en) 2010-05-03 2016-03-29 Pluribus Networks Inc. Methods, systems, and fabrics implementing a distributed network operating system
US9306849B2 (en) 2010-05-03 2016-04-05 Pluribus Networks, Inc. Methods and systems for managing distribute media access control address tables
US9304782B2 (en) 2010-05-03 2016-04-05 Pluribus Networks, Inc. Network switch, systems, and servers implementing boot image delivery
US9319335B1 (en) 2010-12-07 2016-04-19 Pluribus Networks, Inc. Distributed operating system for a layer 2 fabric

Families Citing this family (19)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US8949452B2 (en) * 2005-04-07 2015-02-03 Opanga Networks, Inc. System and method for progressive download with minimal play latency
US8909807B2 (en) * 2005-04-07 2014-12-09 Opanga Networks, Inc. System and method for progressive download using surplus network capacity
US8553720B2 (en) 2006-04-19 2013-10-08 Marvell World Trade Ltd. Adaptive speed control for MAC-PHY interfaces
US8208380B1 (en) * 2006-11-22 2012-06-26 Marvell Israel (M.I.S.L) Ltd. Lossless system support using circular buffer allocation
US8243752B2 (en) * 2007-04-04 2012-08-14 Marvell World Trade Ltd. Long-reach ethernet for 1000BASE-T and 10GBASE-T
US7936669B2 (en) 2008-06-04 2011-05-03 Entropic Communications, Inc. Systems and methods for flow control and quality of service
CN101800706B (en) * 2010-03-25 2015-07-22 中兴通讯股份有限公司 Method and system for obtaining threshold of flow control buffer area
US8539260B2 (en) * 2010-04-05 2013-09-17 Intel Corporation Method, apparatus, and system for enabling platform power states
CN102739510B (en) * 2011-04-15 2016-07-27 中兴通讯股份有限公司 A kind of flow control method and device
CN104641609B (en) * 2012-09-10 2018-03-09 马维尔国际贸易有限公司 Method and apparatus for transmitting packet between the interface control module of Line cards
US8593948B1 (en) * 2012-12-04 2013-11-26 Hitachi, Ltd. Network device and method of controlling network device
US9030936B2 (en) * 2013-06-12 2015-05-12 Intel Corporation Flow control with reduced buffer usage for network devices
US9497125B2 (en) 2013-07-28 2016-11-15 Mellanox Technologies Ltd. Congestion control enforcement in a virtualized environment
JP6287451B2 (en) * 2014-03-26 2018-03-07 富士通株式会社 Data receiving apparatus, data receiving apparatus control method, and data transmitting / receiving system having data transmitting apparatus and data receiving apparatus
US9807024B2 (en) 2015-06-04 2017-10-31 Mellanox Technologies, Ltd. Management of data transmission limits for congestion control
US10009277B2 (en) 2015-08-04 2018-06-26 Mellanox Technologies Tlv Ltd. Backward congestion notification in layer-3 networks
US10237376B2 (en) 2015-09-29 2019-03-19 Mellanox Technologies, Ltd. Hardware-based congestion control for TCP traffic
CN107689923B (en) * 2016-08-04 2021-02-12 华为技术有限公司 Message processing method and router
US11909848B2 (en) 2020-07-09 2024-02-20 Mellanox Technologies, Ltd. Multi-flow compression

Citations (8)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US6170022B1 (en) 1998-04-03 2001-01-02 International Business Machines Corporation Method and system for monitoring and controlling data flow in a network congestion state by changing each calculated pause time by a random amount
US20020087723A1 (en) 2001-01-03 2002-07-04 Robert Williams Method and apparatus for performing priority-based flow control
US20020136163A1 (en) 2000-11-24 2002-09-26 Matsushita Electric Industrial Co., Ltd. Apparatus and method for flow control
US20030046458A1 (en) 2001-08-28 2003-03-06 Kenta Morishima Data transfer equipment and aligner included in data transfer equipment
US20030218977A1 (en) 2002-05-24 2003-11-27 Jie Pan Systems and methods for controlling network-bound traffic
US20040081090A1 (en) 2002-09-09 2004-04-29 Toshihide Hara Congestion controller for ethernet switch
US6745179B2 (en) 2001-10-12 2004-06-01 Shipley Company, L.L.C. Method and system for facilitating viewer navigation through online information relating to chemical products
US6771601B1 (en) 2000-01-31 2004-08-03 International Business Machines Corporation Network switch having source port queuing and methods, systems and computer program products for flow level congestion control suitable for use with a network switch having source port queuing

Family Cites Families (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US6026075A (en) * 1997-02-25 2000-02-15 International Business Machines Corporation Flow control mechanism
DE69923898T2 (en) * 1998-03-14 2006-04-27 Samsung Electronics Co., Ltd., Suwon Apparatus and method for exchanging frame messages of different lengths in a CDMA communication system
US6167029A (en) * 1998-10-13 2000-12-26 Xaqti Corporation System and method for integrated data flow control
JP2003511906A (en) * 1999-10-02 2003-03-25 サムスン エレクトロニクス カンパニー リミテッド Method for fair flow control in packet-switched networks
US6754179B1 (en) * 2000-06-13 2004-06-22 Lsi Logic Corporation Real time control of pause frame transmissions for improved bandwidth utilization
US7062568B1 (en) * 2002-01-31 2006-06-13 Forcelo Networks, Inc. Point-to-point protocol flow control extension

Patent Citations (8)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US6170022B1 (en) 1998-04-03 2001-01-02 International Business Machines Corporation Method and system for monitoring and controlling data flow in a network congestion state by changing each calculated pause time by a random amount
US6771601B1 (en) 2000-01-31 2004-08-03 International Business Machines Corporation Network switch having source port queuing and methods, systems and computer program products for flow level congestion control suitable for use with a network switch having source port queuing
US20020136163A1 (en) 2000-11-24 2002-09-26 Matsushita Electric Industrial Co., Ltd. Apparatus and method for flow control
US20020087723A1 (en) 2001-01-03 2002-07-04 Robert Williams Method and apparatus for performing priority-based flow control
US20030046458A1 (en) 2001-08-28 2003-03-06 Kenta Morishima Data transfer equipment and aligner included in data transfer equipment
US6745179B2 (en) 2001-10-12 2004-06-01 Shipley Company, L.L.C. Method and system for facilitating viewer navigation through online information relating to chemical products
US20030218977A1 (en) 2002-05-24 2003-11-27 Jie Pan Systems and methods for controlling network-bound traffic
US20040081090A1 (en) 2002-09-09 2004-04-29 Toshihide Hara Congestion controller for ethernet switch

Cited By (16)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
EP2048831A1 (en) * 2007-10-12 2009-04-15 Nokia Siemens Networks S.p.A. Method and systems for QoS-aware flow control in communication networks, corresponding network and computer program product
WO2009046869A1 (en) * 2007-10-12 2009-04-16 Nokia Siemens Networks S.P.A. Method and systems for qos-aware flow control in communication networks, corresponding network and computer program product
EP2256640A1 (en) * 2009-05-29 2010-12-01 Oracle America, Inc. Managing traffic on virtualized lanes between a network switch and a virtual machine
CN101902395A (en) * 2009-05-29 2010-12-01 甲骨文美国公司 The system and method for the resource on the management host
US8174984B2 (en) 2009-05-29 2012-05-08 Oracle America, Inc. Managing traffic on virtualized lanes between a network switch and a virtual machine
EP2302847A1 (en) 2009-09-25 2011-03-30 Alcatel Lucent Control of traffic flows with different CoS in a communication network
WO2011036001A1 (en) * 2009-09-25 2011-03-31 Alcatel Lucent Control of traffic flows with different cos in a communications network
US9300576B2 (en) 2010-05-03 2016-03-29 Pluribus Networks Inc. Methods, systems, and fabrics implementing a distributed network operating system
US9042233B2 (en) 2010-05-03 2015-05-26 Pluribus Networks Inc. Method and system for resource coherency and analysis in a network
US9154445B2 (en) 2010-05-03 2015-10-06 Pluribus Networks Inc. Servers, switches, and systems with virtual interface to external network connecting hardware and integrated networking driver
US9160668B2 (en) 2010-05-03 2015-10-13 Pluribus Networks Inc. Servers, switches, and systems with switching module implementing a distributed network operating system
US8767752B1 (en) 2010-05-03 2014-07-01 Pluribus Networks, Inc. Method and system for resource coherency and analysis in a network
US9306849B2 (en) 2010-05-03 2016-04-05 Pluribus Networks, Inc. Methods and systems for managing distribute media access control address tables
US9304782B2 (en) 2010-05-03 2016-04-05 Pluribus Networks, Inc. Network switch, systems, and servers implementing boot image delivery
US9319335B1 (en) 2010-12-07 2016-04-19 Pluribus Networks, Inc. Distributed operating system for a layer 2 fabric
US8891543B1 (en) 2011-05-23 2014-11-18 Pluribus Networks Inc. Method and system for processing packets in a network device

Also Published As

Publication number Publication date
CN101536413A (en) 2009-09-16
EP1882336A4 (en) 2011-08-31
US20060268709A1 (en) 2006-11-30
EP1882336B1 (en) 2013-02-20
WO2006124414A3 (en) 2007-11-22
CN101536413B (en) 2014-06-11
EP1882336A2 (en) 2008-01-30
US7839779B2 (en) 2010-11-23

Similar Documents

Publication Publication Date Title
US7839779B2 (en) Queue aware flow control
US7957285B1 (en) Inter-device flow control
US7903552B2 (en) Directional and priority based flow control mechanism between nodes
EP1280301B1 (en) Method and switch for congestion control
US9094327B2 (en) Prioritization and preemption of data frames over a switching fabric
US8638784B1 (en) Network switch having virtual input queues for flow control
US9992130B2 (en) Systems and methods for flow control and quality of service
US7120113B1 (en) Systems and methods for limiting low priority traffic from blocking high priority traffic
US8059671B2 (en) Switching device
US6295281B1 (en) Symmetric flow control for ethernet full duplex buffered repeater
US8819265B2 (en) Managing flow control buffer
US8018851B1 (en) Flow control for multiport PHY
US7106696B1 (en) Systems and methods for limiting the rates of data to/from a buffer
US20140163810A1 (en) Method for data transmission among ECUs and/or measuring devices
EP2442498B1 (en) Method and device for controlling switching network traffic
US7701949B1 (en) System and method for switching high priority traffic with low latency
US20050138238A1 (en) Flow control interface
JP3995482B2 (en) Router that transmits flow control signals at high speed
Kirstädter et al. A Highly Scalable Switching and Routing Chipset

Legal Events

Date Code Title Description
WWE Wipo information: entry into national phase

Ref document number: 200680010068.1

Country of ref document: CN

121 Ep: the epo has been informed by wipo that ep was designated in this application
WWE Wipo information: entry into national phase

Ref document number: 2006752446

Country of ref document: EP

NENP Non-entry into the national phase

Ref country code: DE

NENP Non-entry into the national phase

Ref country code: RU